321
CONTRACTUAL AGRICULTURE: BETTER PARTNERSHIPS BETWEEN
SMALL FARMERS AND THE BUSINESS SECTOR IN THE SULTANATE OF
OMAN
Badria Hamed Al
Ruqishi a
Tarig Gibreel a
Faical Akaichi b
Lokman Zaibet a
Slim Zekri a
a Department of Natural Resource Economics, College of Agricultural and
Marine Sciences, Sultan Qaboos University, Muscat, Sultanate of Oman
b Department of Rural Economy, Environment and Society, Scotland's Rural
College, Edinburgh, United Kingdom
[email protected] (Corresponding author)
ARTICLE HISTORY:
Received: 28-Feb-2020
Accepted: 04-May-2020
Online Available: 01-Jun-
2020
Keywords:
Contract farming,
Stated preferences,
Transaction costs,
Discrete choice,
Mixed logit model,
Latent class model
ABSTRACT
This study aims to elicit the preferences of Omani small-scale vegetable
farmers towards contracting with different emerging and other potential
business partners along with other contract terms. To accomplish this, a
discrete choice experiment was adopted to elicit farmers' preferences
towards different contract models. The choice data was analyzed using
both latent classes as well as mixed logit models and as a result, three
classes were found to best capture the preferences. Class 1 represents 45%
of the respondents who are more likely to adopt the “multipartite
contract” model. This segment is characterized by a high education level,
older age, and smaller family size. The second class represents 31% of the
sample and tends to adopt an “informal contract” model. This group has a
low experience, larger farm size, and own their private land. The last class
represents 23% of the observations and is in favour of the “centralized
contract” model. The main characteristics of this class are the low
education level, older age, and medium income. Realizing the farmers’
preferences and their characteristics are certainly important in improving
farmers’ participation in the vegetables’ market and gearing the policies
towards food security.
Contribution/ Originality
The significance of this research stems from the genuine lack of research in Oman related to the role of
contractual arrangements involving different governance structures like the form of a true agribusiness
and farmer’s organization. Secondly, we add a new dimension to the literature by analysing farmers’
preferences for different contract models by using a latent class Model and categorizing the choices with
the different farmer’s characteristics.
DOI: 10.18488/journal.1005/2020.10.1/1005.1.321.335
ISSN (P): 2304-1455/ISSN (E):2224-4433
How to cite: Badria Hamed Al Ruqishi, Tarig Gibreel, Faical Akaichi, Lokman Zaibet, and Slim Zekri
(2020). Contractual agriculture: better partnerships between small farmers and the business sector in the
Sultanate of Oman. Asian Journal of Agriculture and Rural Development, 10(1), 321-335.
© 2020 Asian Economic and Social Society. All rights reserved.
Asian Journal of Agriculture and Rural Development Volume 10, Issue 1 (2020): 321-335
http://www.aessweb.com/journals/5005
Asian Journal of Agriculture and Rural Development, 10(1)2020: 321-335
322
1. INTRODUCTION
According to statistics from the Food and Agriculture Organization of the United Nations (FAO),
the Sultanate of Oman relies very heavily on imports to secure roughly 60% of its vegetable’s
needs (FAO, 2017). Food prices fluctuations and the need to be secured for food supplies dictate
the urgent need of having a robust food security policy. Food security-oriented studies in Oman
focused on production efficiency and overlooked the marketing aspect, hence this study has come
to close this gap. This research views farmer’s ability to grow depending upon better access to
high-value chains by participating in different contracts that are patronized by the partnership of
different potential business models. These models have emerged as a result of various measures
that have been taken by the government to ensure food security by focusing on the potential of
vegetable production within Oman. This has resulted in harnessing the private sector by
establishing an agribusiness company that could act as an institutional solution to improve the
efficiency of the sector. Moreover, small-scale farmers started to operate collectively through
farmers’ organizations. Over the years, contract farming has been well-thought-out as a system that
has great potential for providing an approach to integrate small-scale farmers into export,
processing markets and the food security (Dedehouanou et al., 2013; Reardon and Timmer, 2014;
Mishra et al., 2018, Vilbert et al., 2019; Gelli, 2015).
It is one of the institutional options that enable small farmers’ diversification by enhancing their
access to markets, decreasing price risks and transaction costs; contracts that deliver credit, inputs,
technology, extension services, and information, aid farmers increase production efficiency; build
profit-making culture, and increase income and employment (Birthal et al., 2008; Eaton and
Shepherd, 2001; Michelson, 2013; Da Silva and Rankin, 2013). Bellemare and Lim (2018) have
viewed the positive effects of contract farming in enhancing small-scale farmers’ market
participation and its positive impact on their welfare (Wang and Kooten, 2018). Moreover, it
enables farmers to upgrade their production value chain and tap into high-quality markets (Demont
and Rutsaert, 2017). Several authors also have found that engaging in contract farming has
improved small-scale farmers’ income and other studies have examined the impact of contract
farming on high-value crops like vegetables (see Otsuka et al., 2016 and Wang et al., 2014). It has
also been proven that contracts with different patrons such as private companies, processors,
cooperatives, and international companies can improve farmers’ income (Hazell et al., 2007).
Hence, contract farming has been promoted as an institutional innovation and as a critical element
for rural development to improve agricultural performance in developing countries.
Consequently, contract farming can be defined as a profitable relationship between a firm and a
group of farmers. Grosh (1994) noted that contract farming could act as an agro-institution that can
overcome market failure resulting from uncertainty and risk. Most of the studies (Castaneda et al.,
2018; Chamberlain and Anseeuw, 2017; Mishra et al., 2016) rarely focused on the complexity of
the set-up of such arrangements that involve different instruments like collective organization and
agribusiness companies. Different contract models may be perceived differently and these different
views depend on the type of contractor, the price, the length of the contract, and by the
characteristics of the respondents, Bellemare and Lim (2018). Accordingly, contract farming takes
many forms, such as the centralized, informal, and multipartite model (Eaton and Shepherd, 2001).
Contract farming models depend on four main components which are the type of contractor, the
type of product, the intensity of vertical integration, and the involved stakeholder’s number
(Bijman, 2008).
Since one size doesn't fit all (Gregory and Chapman, 2002), different potential contracts were tested
including an agribusiness company, farmers organization, and retailing companies. However,
researchers believe that farmers are still concerned about the contract’s nature and hence there is a
need to elicit their preferences towards those trade-offs. Since farmers weigh the benefits over the
Asian Journal of Agriculture and Rural Development, 10(1)2020: 321-335
323
costs to maximize their utility, farmers face various kinds of opportunity costs before they decide
to engage in a specific contract model. Contract attributes are best analyzed by considering the
trade-offs between returns and risks associated with such a mutual relationship. The terms of the
contract and its levels were selected based on transaction cost theory (Seng, 2017).
Discrete Choice Experiment (DCE) has been used in many areas to measure preferences for choice
alternatives, for instance, DCE has been used to derive farmers’ preferences for contract and market
channel characteristics (Blandon et al., 2009; Schipmann and Qaim, 2011; Landmann et al., 2018;
Fischer and Wollni, 2018; Gelaw et al., 2016; Ochieng et al., 2017; Van den Broeck et al., 2017).
This study uses a DCE that is based on stated preference data. The stated choice method has
initially been developed in marketing research and applied in the contract (Lajili et al., 1997). This
data is collected from the farm households to analyze farmers' preferences for contract farming that
allows controlling the correlation between different attributes for different contract profiles. The
primary purpose of the stated preference approach is to assess these models by eliciting farmer’s
stated preferences towards hypothetical settings (Arouna et al., 2017). This methodology indicates
the maximum utility of a particular contract model by separately estimating the preferences of the
farmers for the pertinent attributes which characterize the contract model. The stated preference
approach is used given the limitation of real contracts on vegetable production in the study area.
Hence, the stated choice method offers an excellent opportunity to estimate the demand for new
potential contracts.
The objective of this paper is to analyze the farmer’s preferences towards different contract
structures in the selected study area in the Sultanate of Oman. This research examined different
contract options using a latent class model to enable policymakers and the emerging agribusinesses
in the vegetable industry to develop a suitable governance structure of contract farming that copes
with the business purpose of different farmers groups.
The paper is organized as follows: Section 2 describes the data collection and data analysis
including the latent class model. Results are displayed in section 3, and the final section provides
the concluding remarks and policy implications.
2. MATERIALS AND METHODS
2.1. Data collection
Farmers’ preferences for which type of contract farming depends on several contract attributes such
as the type of contractor, technical assistance, duration of the contract and pricing mechanism
(Schlecht and Spiller, 2012) For a better qualitative understanding of which behavioral preferences
might affect the valuation of contract characteristics among Omani farmers, focus group
discussions were guided by the theory of transaction costs. Farmers in choosing a contract seek to
lower the transaction costs associated with information, negotiation, and monitoring (North, 1990).
This has helped in evaluating the relevance of specific contract features for farmers and decide how
to design the attributes and which levels to include in the Discrete Choice Experiment. Based on
farmers’ qualitative statements, six attributes were selected. These are shown in Table 1 and vary
systematically in their levels: Type of Partner, Cropping decision rights, quality specifications,
technical assistance, length of the contract, and the price. The levels corresponding to each attribute
of the contract were coded as binary variables except with the price. The level of price specification
was considered based on historical data obtained from Al Batinah farmers’ organization records.
The estimated parameters represent the respondent’s preferences concerning the baseline level.
Indicated (0) represents the reference/baseline for each level.
A seventh variable “None” (NONE) was also considered to estimate respondents’ preferences for
the status-quo option. An example of a card where option four is added to give the farmer the
Asian Journal of Agriculture and Rural Development, 10(1)2020: 321-335
324
freedom and avoid the biases of forcing him/her to choose only from the listed options is displayed
in Appendix A.
Table 1: Summary of contract attributes and levels
No Attributes Levels
1 Type of Partner Retailing firm (0), Farmers’ organization(1), Processing firm (2)
2 Cropping Decision Rights Farmers’ organization(0), Trading Firm (1), Individually (2)
3 Quality specification Variable quality requirement (0), Minimum quality requirement
(1), High-quality standards (2)
4 Technical assistance Provision of technical assistance by the government (0)
Provision of technical assistance by the buyer firm (1)
5 Length of contract
one year (12 months) (0)
one season (9 months) (1)
two years (24 months) (2)
6 Price
Market price (0)
7% less than the market price (1),
7% more than the market price (2),
The data were collected interviewing vegetable producers who were randomly selected from an
already prepared list of 220 common vegetable small-scale farmers’ names obtained from the
Ministry of Agriculture and Fisheries, whose farms with a holding size of fewer than five hectares.
In total 160 farmers were surveyed and interviewed between May and July 2019 in 8 provinces
(wilayat) in Al Dakiliah and Batinah Governorates of the Sultanate of Oman, considered as the
principal vegetable production areas in the country, using a structured questionnaire comprising
two parts. The first part was used to collect the socio-economic and demographic characteristics of
producers. The second part focuses on the choice experiment. To examine the farmer’s preferences
towards different contract models, Ngene software was used to ensure the main effect of interaction
and to generate the efficient total number of choices set combinations. An efficient design of eight
choice sets was generated and broken into three blocks.
Making these cards readable to the farmers with clear levels and attributes, the choice cards are
translated into the Arabic language. Each farmer was asked to complete 8 choice sets and asked to
choose one option out of the three choices presented in every card and a fourth opt-out alternative.
2.2. Data analysis
The data collected were analysed within a random utility framework (McFadden, 1974). Thus, an
individual n presented with j alternatives at a choice occasion t is expected to choose the alternative
that maximizes his/her utility. Following Lancaster’s concept that any product is a bundle of
attributes (Lancaster, 1966), the utility that an individual n derives from the consumption of a
product is assumed to be equal to the sum of his/her marginal utility for each of the product’s
attributes. Therefore, what a farmer derives from a contract farming package is expected to be
equivalent to the summation of the marginal utilities for each of its attributes. Therefore, the
potential value of utility of an individual i’s associated with a contract farming j is defined as 𝑈𝑖𝑗𝑡 ,
which is particular for each jth contract farming alternative a tth choice occasion. It can be written as
a function of two components: an observable systematic component ( 𝑉𝑖𝑗𝑡 ) and a random
component (𝜀𝑖𝑗𝑡 ), which encompasses the unobservable part. Thus, the utility function can be
written as follows:
𝑈𝑖𝑗𝑡 = 𝑉𝑖𝑗𝑡 + 𝜀𝑖𝑗𝑡 …………………… (1)
Where, 𝑉𝑖𝑗𝑡 represents the systematic utility and 𝜀𝑖𝑗𝑡 is assumed to be independent and identically
distributed for all options in each choice set. The systematic utility (𝑉𝑖𝑗𝑡 ) , expressed as a
Asian Journal of Agriculture and Rural Development, 10(1)2020: 321-335
325
generalized regression function (Ben-Akiva and Boccara, 1995) as a function of various
explanatory variables can be written as:
𝑈𝑖𝑗𝑡 = 𝛽𝑋𝑖𝑗𝑡 + 𝜀𝑖𝑗𝑡 …………………… (2)
𝑈𝑖𝑗𝑡 can be expressed by K attributes 𝑋𝑖𝑗𝑡 that are assessed by parameters specific to respondent
choice 𝜷ijt that is not observable and unknown utility parameter by the concerned researchers.
Where β denotes the K×1 vector of unknown marginal utilities that are associated with the Farming
contract attributes Xijt.
The individual preferences that are not observable are to be considered in the stochastic
component𝜀𝑖𝑗𝑡 , assuming an independently and identically distributed extreme value distribution.
Thereby, 𝜷 denotes the vector of unknown parameters that are associated with the farming contract
attributes Xijt. Individuals’ preferences heterogeneity is tested by a model called the Mixed Logit
Model which was suggested by Train (1998). In the Mixed Logit Model, the independence of
Irrelevant Alternatives is relaxed and the following formula defines the conditional choice
probability that individuals choose an alternative j at a choice occasion t as:
𝑝(𝑗|𝑋𝑖𝑡, 𝛽)= ∏ [exp(X′𝑖𝑡,𝑗𝛽𝑞)
∑ exp(𝛽′X′𝑖𝑘𝑡)
𝐽𝑘=1
]𝑇𝑡=1 …………………… (3)
The unconditional choice probability is the expected value of the logit probability integrated over
all possible values of 𝜷 and weighted by the density of:
𝑝(𝑗|𝑋𝑖𝑡, 𝛺) = ∫ 𝑝(𝑗, 𝑋𝑖𝑡, )𝑓(𝛽|𝛺) ⅆ𝛽𝛽
…………………… (4)
As the expression in the above formula does not have one solution, then simulation methods are
created to resolve this issue. The choice probability has been calculated for each draw. Maximum
Likelihood procedures have been used to estimate the simulated log-likelihood (SLL) and are
calculated as:
𝑆𝐿𝐿 = ∑ ∑ 𝑙𝑛𝑇𝑡=1
𝐼𝑖=1 (
1
𝑅 ∑
exp(𝛽′𝑖𝑋𝑖𝑗𝑡)
∑ exp(𝛽′𝑖𝑋𝑖𝑗𝑡)𝑗𝑘=1
𝑅𝑅=1 ) …………………… (5)
Train (2009) has created a new approach called conditional distributions to calculate individual-
specific distributions. The main reason behind this computation is to obtain sufficient information
on the likely position of each respondent. However, using the conditional distributions approach to
evaluate the heterogeneity of respondents’’ preferences is of limited practical use. The estimation
of the Mixed Logit Model has allowed examining that respondents’ attitudes are highly
heterogeneous. Although this model does not explain the source of the heterogeneity it does control
it and considers the attributes variation across respondents.
Therefore, the focus has been shifted to different alternatives termed latent levels (LCM) to analyze
discrete choice analysis and explain the heterogeneity, Greene and Hensher (2003) LCM captures
the source of heterogeneity in the individual’s preferences by assuming a homogenous preferences
pattern within distinct groups. This model is powerful for decision making and policy-relevant
analysis.
The log-likelihood for respondents in the LCM for discrete choice analysis can be described as:
𝑙𝑛𝑙 = ∑ 𝑙𝑛[∑ 𝐻𝑖𝑞(∏ 𝑃𝑖𝑡|𝑞(𝑖)𝑇𝑖𝑡=1 )𝑄
𝑞=1 ]𝑁𝑖=1 …………………… (6)
Asian Journal of Agriculture and Rural Development, 10(1)2020: 321-335
326
Where Hiq denotes prior probability for class q for individual i. for this study, the form of the prior
probability is a multinomial logit
𝐻𝑖𝑞 =Exp(zi′θq)
∑ exp(𝑧′𝑖𝜃𝑞)𝑄𝑞=1
, 𝑞 = 1, … . . , 𝑄, 𝜃𝑄 = 0 …………………… (7)
Where Zi refers to a set of observable characteristics (e,g socio-demographics such as gender, age,
and income) that enter the model for class membership.
Pitq is the chosen probability that individual i, conditional to belonging to class q (q = 1,…q)
chooses alternative i from a particular set j, comprised of j alternatives, in a particular choice
occasion t, and is represented as:
𝑝𝑖𝑡|𝑞(𝑗) =exp(X′𝑖𝑡,𝑗𝛽𝑞)
∑ exp(X′𝑖𝑡,𝑗𝛽𝑞)𝐽𝑗=1
…………………… (8)
Then beta regression analysis uses the computed respondent specific estimates to profile the
members of each class. Two criteria, which are the Consistent Akaike information Criterion
(CAIC) and the Bayesian Information Criterion (BIC), are used to determine the number of classes.
3. RESULTS AND DISCUSSION
The Mixed Logit Model was estimated by using modified Latin Hypercube Sampling draws along
with 1200 simulations, considering repeated choice situations. All the estimations were conducted
using the R-studio software and Stata software. The results displayed in Table 2 show the latent
class model with 3 classes, fit the data better than the conditional logit based on the CAIC and BIC
Values;
Table 2: Information on the converged latent segment models
Number
of Classes
Log likelihood
at convergence (LL)
Number
of parameters (P)
Number
of respondents (N) CAIC BIC
2 -1097.9 31 116 2374.16 2343.16
3 -1052.0 50 116 2392.48 2342.48
4 -1036.3 69 116 2469.60 2400.60
5 -1005.7 88 116 2517.72 2429.72
CAIC (Consistent Akaike Information Criterion) is calculated using: -2 * LL + (ln(N) + 1) * P
BIC (Bayesian Information Criterion) is calculated using: -2 * LL + ln(N) * P
Results derived from the estimation of the Mixed Logit Model and the LCM are presented in Table
3. The results indicate that the estimated marginal utilities are significant. The positive sign of the
coefficient “Opt-out option” suggests a general preference for the opt-out alternative. It is crucial to
recall those respondents who chose “none” for all choices presented have been excluded. Most of
the standard deviation parameters, which show how the valuation of the entire sample spreads
around the estimated means are significant, indicating that the preferences heterogeneity is
maintained among the sampled farmers. The estimated 7% less than the market price coefficient
statistically significant and is negative, indicating that respondents prefer to choose price running
the market rather than the market price, other attributes constant.
3.1. Types of partner
To improve the efficiency of the agricultural sector, the Omani government has decided to establish
an agribusiness company as an institutional solution to link farmers with the market and to ensure
Asian Journal of Agriculture and Rural Development, 10(1)2020: 321-335
327
the food security (Ministry of Agriculture and Fisheries, 2019), it is targeting large scale farming
contracts (Al-Abdali, 2019). Nevertheless, small-scale farmers can integrate with large scale
farming if they operate collectively through farmers’ organizations. However, the forming of such
an organization in Oman is not supported by a legal setting. It is possible to be established as non-
profit organizations, and this legal form constrains it to function as a profit and commercial entity,
(FAO and SQU, 2018).
The results from the Mixed Logit Model, displayed in Table 3 corroborates the existing legislative
status of the farmer’s organization in Oman where that respondents are indifferent about the type of
partner to contract with the farmer’s organization or the trading firm. Nevertheless, respondents in
class 1 and class 2 shows that they are less likely to favor a farmer’s organization compared to the
trading firm. This result is expected, given that the farmer’s organization can lower transaction
costs (Andersson et al., 2015).
Unsurprisingly, the coefficient of the variable “processing firm” is negative and significant at 1%
for class 1, which means that vegetable farmers are more likely to make a contract with retailing
firms such as supermarkets and hypermarkets instead of processing firm. It could be attributed to
the fact that farmers are familiar with the traders than with the processing firm. Moreover, it could
be explained by the farmer’s perception that the processing firms target large scale farms by which
would harm the small scale business farms. Nevertheless, farmers in class 2 are indifferent to adopt
processing firms concerning the retailing partner, but class 3 is more likely to adopt the processing
firm shows a significant level of 5% with a coefficient of 0.713.
3.2. Cropping decision rights
The agricultural commodity market is described by speculative behavior (Tadesse and Gut-
tormsen, 2011). It was expected that farmers would prefer the firm to take the lead of cropping
decision, to avoid unpredicted price fluctuations and the naïve cropping imitation. However, it was
found that overall, farmers are more likely to prefer individual decisions. The results of MLM
displayed in table 3 show that farmers are more likely to decide individually upon the crop’s type
instead of giving the lead to the farmer’s organization. Moreover, farmers are indifferent to the
trading firm and farmer’s organization to decide for them the crop’s type.
Farmer’s organization is the reference line in eliciting the cropping decision to compare the
farmer’s preferences concerning deciding upon the crop type by other partners such as the
Subsidiary trading company or deciding individually. With the latent class model, it shows a
significant sign of 1% for class 1 that is more likely to adopt the decision of trading firm
concerning farmer’s organization. Class 2 and 3 are indifferent to their preferences upon the
decision of cropping. However, with the latent class model, it showed that class 3 are more likely to
adopt the individual decision compared with class 1, while class 2 is the least class that prefers to
go with the individual decision in cropping.
3.3. Quality specification
Quality standards play a major factor in sustaining a healthy contractual relationship (Dolan et al.,
2000; Berdegué et al., 2005). The prior quality specification is crucial to reduce the pertinent
uncertainty related to the demand of the buying firm which results in lower transaction costs
(Goodhue, 2011). Mixed Logit Model in table 3 shows indifferent preferences over the contract that
requires low-quality specification and a contract that specifies the variable quality as a variable.
However, the preferences with high-quality specifications over the variable quality one are highly
significant. With latent class Model, positive and significant results at 5% presented reveal that
farmers in class 2 are more likely to prefer low quality over the variable quality specification and
farmers in class 1 prefer more variable quality specifications.
Asian Journal of Agriculture and Rural Development, 10(1)2020: 321-335
328
In class 2 and 3, they prefer more high-quality specifications with respect to the variable
requirement. This shows that high-quality standards are important for vegetable farmers especially
farmers in class three compared with class 2. On the contrary, farmers in class 1 are indifferent
about the high-quality standard.
3.4. Technical assistance provision
Farmers should have sufficient knowledge and know-how to meet food quality specifications and
enhance productivity. Within the Omani context, farmers can obtain technical assistance from the
government through the Ministry of Agriculture and Fisheries. The statistical analysis has shown a
negative coefficient sign with respect to the technical assistance provided by the government
compared to the buyer firm and this implies that farmers prefer more the assistance to be provided
by the firm instead of the government especially class 1 and class 3. Whereas, class 2 is indifferent
about the source of technical assistance provision. The tendency to favor the provision of technical
assistance by the contracted firms rather than the government can be attributed to the professional
quality approach set by the firm to be followed and found that this professionalism doesn’t exist
with the government (Bellemare, 2010).
According to Masakure and Henson (2005), the technical assistance provided by the buying firm
incentivizes farmers to enhance their farm performance and farmers’ accumulative farming
experience.
3.5. Duration of the contract
Overall, 76% of farmers are indifferent between the three proposed durations of contract. This
implies that farmers prefer the short-duration contract of 9 months compared to 12 months (the
baseline). This tendency towards shorter contracts could be attributed to the fact of farmer’s
uncertainty and risk aversion and could be also explained by the climate adaptability to grow
vegetables during one season only. Short contracts allow the farmers to abrogate the contract if the
partner does not adhere to an agreement. At the same time, other farmers are indifferent about the
length of the contract. The variable “contract of 9 months” coefficient is significant for latent class
3 where the marginal effects are estimated at 0.970 and represent only 23% of sampled farmers.
The other level of contract length “contract 24 months” in the Mixed logit model reveals farmer’s
indifferences about this contract with respect to “contract of 12 months”. This confirms the
preferences by latent class model, the three classes are indifferent about the length of the contract of
24 months with respect to 12 months.
3.6. Price
Price is considered as the first term discussed between parties in contracts( Hernández et al., 2007).
The coefficient of “7% less than market price” is significant and negative. Not surprising, this
revealed that contracts with market prices increase the probability for farmers to adopt a contract.
Market prices guarantee the market for their production (Michelson, 2012). Besides, the market
price reduces the uncertainty associated with fluctuated prices. The same attribute’s coefficient is
negative and significant with latent class 2 that reveals the preference of market price. The level of
7% more than market price shows a coefficient is insignificant with the Latent class model and
Mixed Logit Model, which indicates their indifferences about this level.
Asian Journal of Agriculture and Rural Development, 10(1)2020: 321-335
329
Table 3: Estimated farmers’ preferences
Mixed logit model
(MLM) Latent class model (LCM)
Variables Mean standard
deviations Class 1 Class 2 Class 3
Farmer organization 0.083 -- -0.012 -0.362 0.258
Processing firm -0.492** 1.109*** -0.812*** 0.008 0.713**
Cropping decision by a
Trading firm -0.091 1.142*** 0.758*** 0.222 0.030
Individual cropping decision 0.433* 0.876*** 0.064 -0.262 0.367
Low quality 0.143 0.846** -0.423 0.852** -0.662
High quality 0.589*** 0.648*** 0.370 0.409* 0.824***
Technical assistance by
government -0.209 -- -0.558** 0.201 -0.730***
Contract 9 months 0.098 -- 0.390 0.122 0.970**
Contract 24 months -0.086 -- 0.117 0.144 0.307
7% less than market price -0.538*** -- -0.333 -0.572** -0.506
7% more than the market
price -0.096 -- 0.300 0.046 -0.758
Opt-out option 3.268*** 12.147*** 1.545*** -1.665*** -1.985**
Class shares 1
0.45 0.31 0.23
***, **, and * denotes significant at 1%, 5%, and 10% level of significance respectively.
Table 4: Regression results based on marginal effects
Variables Class 1 Class 2 Class 3
Coef. Std.Err Coef. Std.Err Coef. Std.Err
Region -1.937*** 0.049 0.461*** 0.093 10.4*** 0.324
Gender -0.146** 0.067 -0.046 0.107 0.006 0.008
Low education 0.623*** 0.126 -0.908*** 0.136 0.313*** 0.038
High education 0.706*** 0.057 0.371*** 0.097 -1.681*** 0.008
Age 0.004* 0.002 0.008*** 0.003 0.001* 0.000
Family size -0.012* 0.007 -0.033*** 0.011 -0.002** 0.001
Main job 0.154*** 0.045 0.321*** 0.069 0.015 0.011
Low income 1.093*** 0.057 0.015 0.099 -1.983*** 0.012
High income -0.273*** 0.058 0.204* 0.118 -0.005 0.011
Marketing experience -0.004** 0.002 -0.008*** 0.003 0.000 0.000
Land ownership 0.068 0.053 0.085 0.098 -0.015* 0.008
Medium operated land -0.051 -0.850 -0.017 -0.170 -0.005 -0.690
Large operated land 0.101* 1.880 0.209*** 2.440 0.015 1.370
Inherited land 0.138** 2.110 0.109 1.030 0.002 0.180
Rented land 0.393*** 0.105 0.669*** 0.178 0.033 0.025
Shared land 0.371*** 0.104 0.589*** 0.202 0.029 0.022
Government land 0.613*** 14.280 -0.128* -1.790 -0.715*** -80.110
Low cultivated vegetables 0.038 0.044 0.014 0.074 -0.006 0.008
High cultivated vegetables 0.011 0.077 -0.057 0.160 -0.035 0.017
Constant -0.168 0.171 -1.401*** 0.295 -9.642 0.327
Class share 0.45 0.31 0.23
Log-likelihood final 221.52 141.26 833.42
Numberof observation 111 111 111
Wald test value 11763.3 243.88 729812.5
P-value <0.01 <0.01 <0.01
***, **, and * denotes significant at 1%, 5%, and 10% level of significance respectively.
Asian Journal of Agriculture and Rural Development, 10(1)2020: 321-335
330
3.7. Heterogeneity
To gain more information about the hypothesis of this research that one size of the contract does
not fit all, it was necessary to identify different segments with similar preferences. The estimated
standard deviations are significant and indicating the presence of heterogeneity which corroborates
the hypothesis of this research. Moreover, identifying the segments of farmers with homogeneous
preferences would help to efficiently target them using strategies tailored to their specific needs.
Hence, the abovementioned three Latent class models described how farmers’ responses varied
across different segments. In this section, the features that describe the preferences of those groups
are profiled and labeled according to the characteristics of the farmers by using the results of Beta
Regressions displayed in table 4. The variables are described in Appendix B. Beta Regression
analysis is used to determine the relationship between the probability of a farmer to belong to one
of the three classes and the independent variables considered in the regression. This can help design
and tailor the policy that fits the needs of the members of each class and identify the contract model
type which fits the group. Each described class is provided with a title that is compatible with the
contract type. The Probability Scale of the socio-demographic variables presented in Table 4 helps
understand how different classes vary against these indicators and what are the socio-demographic
characteristics of each class.
Table 5: Summary of the preferred contracts by the members of each class
Attributes Class 1
“Multipartite model”
Class 2
“Informal model”
Class 3
“Centralized model”
Type of Partner Retailing Firms indifferent Processing firm
Cropping Decision Trading firm Indifferent Indifferent
Quality specification Indifferent low quality High quality
Technical assistance Trading firm Indifferent Trading firm
Length of contract Indifferent Indifferent 9 months
Price Indifferent Market price Indifferent
Class 1 “Multipartite Model”
This class represents 45% of the entire respondents, and It is the class whose members are
indifferent about farmers’ organization. They are also unwilling to contract with processing firms
and prefer retailing firms instead. To cope with uncertainty and to reduce transaction costs, farmers
in this class are willing to be led by the trading firm in deciding upon the crop types. This group is
also indifferent about the quality specification. This class is more inclined towards seeking the
technical assistance provided by the buying firm. This class preferences match with the attributes of
the multipartite model in terms of the needed coordination and the joint venture business model.
The results displayed in Table 5 show that the Dakhiliya region farmers are more in class 1
compared to the Batinah farmers. Moreover, in this class older people, large family size, and those
whose main job is farming are presented comparing with other demographic segments. In addition,
this class indicates that those of low income are well ahead of those characterized as high income.
Also, this class includes more of those of high marketing experience and those who own highly
operated farms. Farmers who attained land from the government and who inherited their lands are
more in this class.
Class 2 “Informal contract model” This class represents 31 % of the respondents. This type of contract is characterized by its transitory
nature where the farmers are not usually engaged in directed farming.
It is characterized by their indifferent preferences between retailing firms, the farmers’ organization
and the processing firm. They are also indifferent about the decision of selecting the crop. Farmers
in this class prefer low-quality standards. With regards to the provided technical assistance, results
in Table 3 show that farmers are indifferent about it whether to be provided by the government or
the buying firm. The length of the contract is also not crucial to the farmers as the price.
Asian Journal of Agriculture and Rural Development, 10(1)2020: 321-335
331
Unsurprisingly, Farmers prefer the “market price” more than the 7% lower price”. The class
preferences match the informal contract model type. From Table 4, it can be noticed that this
member of class 2 is more likely to be located in the Batinah region compared to Dakhliah. The
same segment is more likely to have highly educated farmers and less likely to have a large family
size. In this segment, the farmers whose main job is farming is proportionally higher compared to
other classes. The farm is larger compared to other classes’ farms. Farm ownership indicates that
this group has less governmental land and more rented or shared private land. The marketing
experience of this group shows the negative value and this indicates the low experience the farmers
have in terms of several years.
Class 3 “Centralized contracting model”
This model main characteristics are directed contract farming where there is a commitment from
the Private corporate sector to provide technical assistance to farmers to meet the high-quality
standards. This group represents 23% of the observations. Farmers are indifferent about contracting
with the farmers’ organization or the retailing firm. However, the segment has stronger preferences
to contract with the processing firm. The members of this class are indifferent about the production
decision. The segment also prefers the high-quality standards significantly with respect to the
variable quality specification. The technical assistance provided by the buying firm is much more
preferred compared to the one provided by the government. The members of this segment prefer
the shortest contract of 9 months. These class preferences attribute match more with the centralized
contracting model type, quality specification and technical assistance. Farmers of this class tend to
have low education levels, old age, and medium-income and tend to own their land.
4. CONCLUSIONS
Contract farming is important to solve farmers' marketing problems in Oman. However, there exists
no “one size fits all” contract; farmers exhibit different preferences for partners’ types and terms of
the contract. This study showed that 76% of the farmers prefer contracting with retailing firms and
farmer organizations while 24% of farmers believe that the centralized contract model is their
choice and the processing firm which is supported by the government is the main partner for them.
Policymakers should ensure prospected policies are matched with the farmer’s preferences in
setting legal profit-oriented producers’ organizations.
Funding: This study received a financial support.
Competing Interests: The authors declared that they have no conflict of interests.
Contributors/Acknowledgement: This project is supported by the internal grant (IG/AGR/ECON/18/01)
and from Sultan Qaboos University bench fees given to PhDs students. This work also would not have
been finalized without the assistance of Dr. Faical Akaichi from Scotland’s Rural College at Edinburgh
University. Views and opinions expressed in this study are the views and opinions of the authors, Asian Journal of
Agriculture and Rural Development shall not be responsible or answerable for any loss, damage or
liability, etc. caused in relation to/arising out of the use of the content.
References
Andersson, C. I., Change, C. G., Rao, E. J., & Qaim, M. (2015). Following up on smallholder
farmers and supermarkets in Kenya. American Journal of Agricultural Economics, 97(4),
1247-1266. http://doi.org/10.1093/ajae/aav006.
Arouna, A., Adegbola, P. Y., Zossou, R. C., Babatunde, R., & Diagne, A. (2017). Contract farming
preferences of smallholder rice producers in Benin: a stated choice model using mixed logit.
Tropicultura, 35(4), 180-191.
Al-Abdali, S. (2019). April 17, Sultanate of Oman, Muscat, food security company. Personal
interview about the farming contract.
Asian Journal of Agriculture and Rural Development, 10(1)2020: 321-335
332
Bellemare, M. F., & Bloem, J. R. (2018). Does contract farming improve welfare? A review. World
Development, 112, 259-271.
Bellemare, M. F. (2010). Agricultural extension and imperfect supervision in contract farming:
evidence from Madagascar. Agricultural Economics, 41, 507-517.
http://doi.org/10.1111/j.1574-0862.2010.00462.x.
Ben-Akiva, M., & Boccara, B. (1995). Discrete choice models with latent choice sets. International
Journal of Research in Marketing, 12(1), 9-24. http://doi.org/10.1016/0167-8116(95)00002-j
Berdegué, J. A., Balsevich, F., Flores, L., & Reardon, T. (2005). Central American supermarkets’
private standards of quality and safety in procurement of fresh fruits and vegetables. Food
Policy, 30, 254-269. http://doi.org/10.1016/j.foodpol.2005.05.003.
Blandon, J., Henson, S., & Cranfield, J. (2009). Small‐scale farmer participation in new agri‐food
supply chains: Case of the supermarket supply chain for fruit and vegetables in Honduras.
Journal of International Development: The Journal of the Development Studies Association,
21(7), 971-984.
Bijman, J. (2008). Contract farming in developing countries. An overview. Wageningen:
Wageningen University. The Netherlands.
Birthal, P. S., Jha, A. K., Tiongco, M. M., & Narrod, C. (2008). Improving farm-to-market linkages
through contract farming a case study of smallholder dairying in India. IFPRI Discussion
Paper 00814 http://www.environmentportal.in/files/IFPRIDP00814.pdf.
Chamberlain, W. O., & Anseeuw, W. (2017). Contract farming as part of a multi-instrument
inclusive business structure: a theoretical analysis. Agrekon, 56(2), 158–172.
http://doi.org/10.1080/03031853.2017.1297725.
Castaneda, A., Doan, D., Newhouse, D., Nguyen, M. C., Uematsu, H., & Azevedo, J. P. (2018). A
new profile of the global poor. World Development, 101, 250-267.
http://doi.org/10.1016/j.worlddev.2017.08.002
Da Silva, C. A., & Rankin, M. (2013). Contract farming for inclusive market access. FAO, Rome,
Italy. Page 23.
Dedehouanou, S. F. A., Swinnen, J., & Maertens, M. (2013). Does contracting make farmers
happy? Evidence from Senegal. Review of Income and Wealth, 59, 138-160.
Demont, M., & Rutsaert, P. (2017). Restructuring the Vietnamese rice sector: towards increasing
sustainability. Sustainability, 9(2), 1-15. http://doi.org/10.3390/su9020325.
Dolan, C., & Humphrey, J. (2000). governance and trade in fresh vegetables: the impact of UK
supermarkets on the African horticulture industry. Journal of Development Studies, 37, 147-
176.
Eaton, C., & Shepherd, A. W. (2001). Contract farming. Partnerships for growth. Economics.
Journal of African Economies, 3(2), 231-261.
Fischer, S., & Wollni, M. (2018). The role of farmers’ trust, risk and time preferences for contract
choices: Experimental evidence from the Ghanaian pineapple sector. Food policy, 81, 67-81.
Food and Agriculture organization and Sultan Qaboos University (2018). Strengthening national
capacities of producer organizations in the sultanate of Oman. Muscat.
http://www.fao.org/3/ca8795en/CA8795EN.pdf.
Food and Agriculture Organization of the United Nations. (2017). FAOSTATA statistical database.
http://www.fao.org/faostat/en/#data/FBS.
Gelli, A., Hawkes, C., Donovan, J., Harris, J., Allen, S., de Brauw, A., Henson, S., Johnson, N.
Garrett, J., & Ryckembusch, D. (2015). Value chains for Nutrition. Discussion Paper 01413.
International Food Policy Research Institute.
Goodhue, R. E. (2011). Food quality: the design of incentive contracts. Annual Review of Resource
Economics, 3, 119-149.
Gregory, G., & Chapman, C. (2002). Differentiated instructional strategies: One size doesn't fit all.
Thousand Oaks, Calif: Corwin Press.
Gelaw, F., Speelman, S., & Van Huylenbroeck, G. (2016). Farmers’ marketing preferences in local
coffee markets: Evidence from a choice experiment in Ethiopia. Food policy, 61, 92-102.
Asian Journal of Agriculture and Rural Development, 10(1)2020: 321-335
333
Greene, W. H., & Hensher, D. A. (2003). A latent class model for discrete choice analysis:
contrasts with mixed logit. Transportation Research Part B: Methodological, 37(8), 681-
698.
Grosh, B. (1994). Contract farming in Africa: an application of the new institutional economics.
Journal of African Economies, 3(2), 231-261.
Hazell, P. B., Poulton, C., Wiggins, S., & Dorward, A. (2007). The future of small farms for
poverty reduction and growth (Vol. 42). Intl Food Policy Res Inst.
Hernández, R., Reardon, T., & Berdegué, J. (2007). Supermarkets, wholesalers, and tomato
growers in Guatemala. Agricultural Economics 36(3), 281-290.
http://doi.org/10.1111/j.1574- 0862.2007.00206.x.
Lajili, K., Barry, P. J., Sonka, S. T., & Mahoney, J. T. (1997). Farmers' preferences for crop
contracts. Journal of Agricultural and Resource Economics, 22(2), 264-280.
Lancaster, K. J. (1966). A new approach to consumer theory. Journal of Political Economy, 74(2),
132-157.
Landmann, D., Feil, J. H., Lagerkvist, C. J., & Otter, V. (2018). Designing capacity development
activities of small-scale farmers in developing countries based on discrete choice
experiments (No. 277738). International Association of Agricultural Economists (pp.7-10).
Masakure, O., & Henson, S. (2005). Why do small-scale producers choose to produce under
contract? lessons from nontraditional vegetable exports from Zimbabwe. World
Development, 33(10), 1721-1733.
McFadden, D. (1974). Conditional logit analysis of qualitative choice behavior. Frontiers in
Econometrics, 105-142.
Michelson, H. (2013). Small farmers, NGOs, and a Walmart world: Welfare effects of
supermarkets operating in Nicaragua. American Journal of Agricultural Economics, 95(3),
628–649.
Michelson, H., Reardon, T., & Perez, F., 2012. Small farmers and big retail: Trade-offs of
supplying supermarkets in Nicaragua. World Development, 40(2), 342–354.
http://doi.org/10.1016/j.worlddev.2011.07.013.
Ministry of Agriculture and Fisheries. (2019). Prospects and opportunities for Agriculture fish and
food investment. Sultanate of Oman. Muscat.
Mishra, A. K., Kumar, A., Joshi, P. K., & D'souza, A. (2016). Impact of contracts in high yielding
varieties seed production on profits and yield: The case of Nepal. Food Policy, 62, 110-121.
Mishra, A. K., Kumar, A., Joshi, P. K., & D'Souza, A. (2018). Impact of contract farming on yield,
costs and profitability in low‐value crop: evidence from a low‐income country. Australian
Journal of Agricultural and Resource Economics, 62(4), 589-607.
North, D. C. (1990). A transaction cost theory of politics. Journal of Theoretical Politics, 2(4), 355-
367.
Ochieng, D. O., Veettil, P. C., & Qaim, M. (2017). Farmers’ preferences for supermarket contracts
in Kenya. Food Policy, 68, 100-111.
Otsuka, K., Nakano, Y., & Takahashi, K. (2016). Contract farming in developed and developing
countries. Annual Review of Resource Economics, 8, 353-376.
Reardon, T., & Timmer, C. P. (2014). Five inter-linked transformations in the Asian agrifood
economy: Food security implications. Global Food Security, 3(2), 108-117.
http://doi.org/10.1016/j.gfs.2014.02.001.
Schlecht, S., & Spiller, A. (2012). A latent class cluster analysis of farmers’ attitudes towards
contract design in the dairy industry. Agribusiness, 28, 121-134.
Seng, W. K. (2017). The role and importance of the transactions costs theory in agricultural
contracting area: an appraisal of selected empirical studies. Journal of Business and
Management, 19(1), 79-89.
Schipmann, C., & Qaim, M. (2011). Supply chain differentiation, contract agriculture, and farmers’
marketing preferences: The case of sweet pepper in Thailand. Food policy, 36(5), 667-677.
Tadesse, G., & Guttormsen, A. G. (2011). The behavior of commodity prices in Ethiopia.
Agricultural Economics, 42, 87-97.
Asian Journal of Agriculture and Rural Development, 10(1)2020: 321-335
334
Train, K. (1998). Recreation demand models with taste variation over people. Land Economics,
74(2), 230-239.
Train, K. E. (2009). Discrete choice methods with simulation. Cambridge university press.
Retrieved from
https://scholar.google.com/scholar?cluster=16561045760721102731&hl=en&as_sdt=0,5.
Van den Broeck, G., Vlaeminck, P., Raymaekers, K., Velde, K. V., Vranken, L., & Maertens, M.
(2017). Rice farmers' preferences for fairtrade contracting in Benin: Evidence from a discrete
choice experiment. Journal of Cleaner Production, 165, 846-854.
Vilbert, V. V., Rajat, P., Shannon, M., Hagerman, C. G., & Robert, A. K. (2019). Achieving
Sustainable Development Goals in the global food sector: A systematic literature review to
examine small farmers engagement in contract farming, Business Strategy & Development,
2(4), 276-289.
Wang, H. H., Wang, Y., & Delgado, M. S. (2014). The transition to modern agriculture: Contract
farming in developing economies. American Journal of Agricultural Economics, 96(5),
1257-1271.
Wang, S., & van Kooten, G. C. (2018). Forestry and the new institutional economics: An
application of contract theory to forest silvicultural investment. Routledge.
Appendices
Appendix A: Choice cards “Sample”
Attributes Option 1 Option 2 Option 3 Option 4
Type of Partner Farmer’s
organization Processing Firm Retailing firm
None
of
these
Cropping Decision Rights Farmer’s
organization
Farmer’s
organization Individually
Quality specification
Variable
quality
requirement
Highly quality
standards
Minimum quality
requirement
Technical assistance Provision by
the buyer firm
Provision by the
buyer firm
Provision by the
government
Length of contract Duration of
contract 1 year
Duration of
contract 2 years
Duration of contract
one season
Price Market price 7% more than
market price
7% less than market
price
please put tick on your
choice
Appendix B: Description of beta regression variables
Variable Description
Region 0 = if the farmer is from the Al Batinah Region and, 1= if the farmer is from
AlDakiliha ( interior region )
Gender 0 = if farmers is female, 1 = if the farmer is male.
Low education This variable is coded if 1 the farmer education attainment level is primary,
intermediary, otherwise is 0
Medium education This variable is coded 1 if the farmer education attainment level is
secondary or diploma, otherwise is 0
High education This variable is coded 1 if the farmer reveals a high education of university
Asian Journal of Agriculture and Rural Development, 10(1)2020: 321-335
335
degrees and above.
Age This variable is continuous and indicates the farmer’s age in years.
Family size This variable is continuous and represents how many children the farmer
has.
Main job This variable is coded as = 1 if the main job is farming, otherwise is = 0.
Low income This variable is coded as 1 if the income range between 100 to 600,
otherwise is 0
Medium income This variable is coded as 1 if the income range between 600- 2000,
otherwise is 0
High income This variable is coded as 1 if the farmer‘s income is higher than 2000 OMR.
Marketing
experience
This variable is continuous to represent how many experience years the
farmer has in marketing.
Land ownership This variable is coded if farmers own his land as 1, otherwise is coded as 0
Low operated land This variable is coded if the operated land range is less than 5.0, otherwise
is = 0
Medium operated
land
This variable is coded if the operated land range is between 5.0 –19.9
Feddan, otherwise is = 0
Large operated land This variable is coded as 1 if the operated land Large >= 20.0 Feddan,
otherwise is = 0
Inherited land This variable is coded as 1 if the land is inherited, otherwise is 0
Rented land This variable is coded as 1 if the land is rented, otherwise is 0
Shared land This variable is coded as 1 if the land is shared with another partner,
otherwise is 0
Purchased land This variable is coded as 1 if the land is purchased, otherwise is 0
Government land This variable is coded as 1 if the government grants the land, otherwise is 0
Low cultivated
vegetables This variable is coded as 1 if the cultivated vegetables is < 5.0 ton
High cultivated
vegetable This variable is coded as 1 if the cultivated vegetables is < 20 ton